CO J = 2 1 MAPS OF BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS

Transcription

1 The Astronomical Journal, 129: , 2005 January # The American Astronomical Society. All rights reserved. Printed in U.S.A. CO J = 2 1 MAPS OF BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS Y. Wu, 1 Q. Zhang, 2 H. Chen, 3 C. Yang, 4 Y. Wei, 1 and P. T. P. Ho 2 Received 2003 December 23; accepted 2004 September 28 ABSTRACT Using the eight-beam array receiver of the NRAO 12 m telescope, we mapped the CO J ¼ 2 1emissiontoward 11 high-mass star-forming regions. In the sample, four are previously detected outflows in the CO J ¼ 1 0 line, and seven are outflow candidates. A total of six bipolar outflows were identified in the CO J ¼ 2 1 line. For the remaining five sources, including one previously detected bipolar outflow, the CO J ¼ 2 1 emission shows multiple velocity components. Therefore, high-velocity line wings or bipolar structure cannot be identified. The CO J ¼ 2 1 spectra of the four of the nonbipolar outflow sources exhibit broad-line emission due to the blending of weak velocity components. The complex CO spectra underscore the importance of large-scale mapping in identifying outflows. Compared with the outflows detected with the CO J ¼ 1 0 line, the CO J ¼ 2 1 outflows often have broader line wings and smaller spatial extents, indicating that the high-velocity gas measured with the CO J ¼ 2 1 line arises from warm regions closer to the central source. The masses in the outflows range from a few to 60 M. The linear momenta in the outflows are as large as a few hundred M km s 1. Both parameters are much larger than the typical values in low-mass outflows. The average dynamic timescale of the outflows is 2 ; 10 4 yr. The driving sources of the bipolar outflows are also identified. All bipolar outflows detected have a near-infrared source, except for IRAS , and all are associated with centimeter or millimeter continuum emission, except for IRAS We investigated the correlation between the outflow parameters and the properties of the driving source. The outflow luminosity and mechanical force correlate with the bolometric luminosity of the star. However, the mechanical force required to drive a CO outflow is more than an order of magnitude higher than the radiation pressure from the star. We reexamined the relation between the mass entrainment rate of the outflows and the bolometric luminosity of the central source with an up-to-date sample. Results show that the mass outflow rate increases with increasing bolometric luminosity, suggesting that the mass outflow rate is related to the luminosity of the central source. Key words: ISM: jets and outflows ISM: kinematics and dynamics ISM: molecules masers stars: formation 1. INTRODUCTION Molecular outflow toward a high-mass young stellar object (YSO) was first detected in Orion A (Zuckerman et al. 1976; Kwan & Scoville 1976). In an early survey by Bally & Lada (1983), most of the objects were high-mass star-forming regions. However, the study of low-mass outflows advanced much more rapidly after the extended and collimated outflow in L1551 was detected (Snell et al. 1980). Until 1999, less than 30 massive outflows were cataloged (Churchwell 1999) when the total number of molecular outflows exceeded 200 (Wu et al.1996). Efforts have been made to search for high-mass outflows in recent years (Beuther et al. 2002a; Zhang et al. 2001; Ridge & Moore 2001; Shepherd & Churchwell 1996a, 1996b). A large number of new outflows have been uncovered. Nevertheless, the search for outflows from massive YSOs is hindered by both the complexity of and the relatively large distances to these sources. In particular, massive YSOs are usually associated with a cluster of lower mass YSOs. The possibility of multiple outflow lobes complicates the identification. 1 Department of Astronomy, CAS-PKU Joint Beijing Astrophysics Center, Peking University, Beijing , China; yfwu@bac.pku.edu.cn. 2 Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138; qzhang@cfa.harvard.edu. 3 Department of Astronomy and Steward Observatory, University of Arizona, Tucson, AZ 85721; hua_chen@net.com. 4 Department of Astronomy, Columbia University, New York, NY 10027; cyyang@astro.columbia.edu. 330 Outflows are usually searched for with the easily excited CO J ¼ 1 0 line. However, by observing lines at higher frequencies, one can attain higher angular resolution, which is critical to reduce the beam dilution and to identify relatively compact outflows. In addition, higher J transitions are relatively more sensitive to hot gas when the high-velocity gas is optically thin. Such hot gas are more likely to be physically associated with YSOs. In this paper we present observations of the CO J ¼ 2 1 line toward 11 high-mass star-forming regions. Four (OMC 2, GGD 27, G , and IRAS ) were selected from outflows measured with the CO J ¼ 1 0 line (Wu et al and references therein). The remaining were selected from outflow candidates detected in previous CO J ¼ 1 0 surveys of massive star-forming regions or water maser sources (Wu et al. 1998a, 1998b). The sources and their references are listed in Table 1. We describe the observations in x 2. The results and discussions of the outflows and their driving sources are given in xx 3 and 4. We summarize in x OBSERVATIONS Observations of the CO J ¼ 2 1 line were made with the 12 m telescope of NRAO 5 during 1998 February The beam size at this wavelength ( GHz) is about 29 00, 5 The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.

2 BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS 331 TABLE 1 Observed Sources Source Name (1) R.A. ( B1950) (2) Decl. ( B1950) (3) D ( kpc) (4) H 2 O (5) CO J = 1 0 a (6) References (7) V LSR (km s 1 ) (8) Bipolar Outflow (9) OMC Y O 1, Y GGD Y O 1, Y IRAS Y C N G Y O 1, N IRAS Y N S Y O Y IRAS Y C Y IRAS Y O 1, Y IRAS N C N IRAS N C N IRAS Y O Y Note. Units of right ascension are hours, minutes, and seconds, and units of declination are degrees, arcminutes, and arcseconds. a O and C indicate CO J ¼ 1 0 outflow and outflow candidate, respectively. References. (1) Wu et al. 1996; (2) Fischer et al. 1985; (3) Yamashita et al. 1989; (4) Wu et al. 1998a; (5) Dent et al. 1985; (6) Comoretto et al. 1990; (7) Ho et al. 1982; (8) Wu et al. 1998b. and the pointing accuracy is The aperture efficiency mb is 32%. The eight beam array receiver was used, with system temperatures of about K. The hybrid spectrometer of 1536 channels was employed and was split into eight sections during the observation. Each section has 192 channels with a total bandwidth of 150 MHz. The equivalent velocity resolutionis1.02kms 1. The absolute position-switching mode was used. To avoid CO emission in the Galactic plane, we chose a reference position a few degrees off in Galactic latitude from the target source. Usually a clean CO offset position was found within two to three trials. The map sizes were typically about 5 0 ; 5 0 and were extended to cover the line wing emissions. The grid spacing of the map is The integration time is 2 minutes for each position. The antenna temperature TA has been corrected for atmospheric attenuation, radiation loss, and rearward and forward scattering and spillover. Data were reduced with the CLASS software. To further identify the excitation sources of outflows, we observed 7 mm continuum emission in OMC 2 with the Very Large Array (VLA) of NRAO on 2001 September 30 in the compact D configuration. The pointing center of the observations was R:A:(B1950) ¼ 05 h 32 m 59: s 58, decl:(b1950) ¼ B4. We used the quasar to calibrate the time variations of gains. Taking advantage of the angular proximity of the calibrator and the target source, we used the fast-switching mode with a calibration duty cycle of 2 minutes. The rms in the image is about 0.2 mjy. 3. RESULTS The results of outflow detection are summarized in Table 1. Columns (1) (3) list the source names and the equatorial coordinates. Column (4) lists the distances, some of which are quoted from references, whereas others are calculated using the Galactic rotation model ( Wouterloot & Brand 1989). Column (5) presents H 2 O maser detection. Column (6) gives the results of the previous search for outflows in the CO J ¼ 1 0line.Column (7) lists the references. Column (8) presents the central velocities (V LSR )ofthecoj ¼ 2 1 line. Column (9) gives the outflow detection in CO J ¼ 2 1. To identify CO outflows, we first examine whether there are high-velocity wings in the CO J ¼ 2 1 spectra near the source positions and whether the intensities of the wing emission decrease smoothly toward the edge of the mapped region. In cases in which multiple velocity components are present, the wing emissions are likely from additional cloud components rather than from the high-velocity gas. The data are further examined in position-velocity (PV) diagrams for intervening cloud components. If confirmed, they should be excluded from the highvelocity emission. As a result, we found bipolar outflows in six sources. In the other five sources, including one previously identified as a bipolar outflow in CO J ¼ 1 0, bipolar outflows cannot be confidently identified because of the confusion of multiple components. Sources are described in two groups in xx 3.1 and 3.2. To analyze the properties of the YSOs in the region, we present in Table 2 the properties of the IRAS sources. The IRAS name, position, and the two color indexes are given in columns (2) (6). The T d in column (7) is the color temperature deduced from the 100 and 60 m flux densities. The infrared luminosity L FIR in column (8) is calculated using L FIR L m ¼ D 2 1:75 F 12 0:79 þ F 25 2 þ F 60 3:9 þ F 100 9:9 (Casoli et al. 1986). To identify driving sources, we make use of the Two Micron All Sky Survey (2MASS) and other nearinfrared data. The 7 mm map of OMC 2 from the VLA is presented. The possible driving sources are presented in Table Sources with Outflows For each outflow source we present typical CO spectra with the blue- and redshifted emission. We determine the velocity ranges of high-velocity gas emission from PV diagrams. The PV diagram was drawn with a cut along the direction where the CO lines have the largest wings. We choose the beginning of the high-velocity wings at the value at which a velocity gradient clearly begins and is spatially extended. With PV diagrams, we can also identify additional velocity components (see x 3.2). For OMC 2 and GGD 27, we adopted the velocity ranges of CO J ¼ 1 0 that have been confirmed by the PV diagram (Fischer et al. 1985; Yamashita et al. 1989; note that for GGD 27 there is a2kms 1 systematic difference between the measured V LSR of Yamashita et al. and this work). For IRAS , the CO J ¼ 1 0 outflow was measured by Rodríguez et al. (1980a). They adopted the highvelocity ranges as 60 to 21 km s 1 and 1to14kms 1 for

3 332 WU ET AL. TABLE 2 Infrared Parameters Name (1) IRAS Source (2) Position a (s) (3) Position a (arcsec) (4) log (F 25 /F 12 ) (5) log (F 60 /F 12 ) (6) T d (K) (7) L FIR (L ) (8) OMC ; b 3.0 ; 10 2 GGD ; 10 4 IRAS ; 10 5 G ; ; 10 4 IRAS ; ; 10 5 S ; 10 4 IRAS ; 10 4 IRAS ; 10 4 IRAS ; 10 4 IRAS ; 10 5 IRAS ; 10 4 a For an IRAS source, HHMMSDDAA, its position is R:A: ¼ HH h MM m ss: s s, decl: ¼ DD AA 0 aa 00, where is ss.s is the value in column (3) and aa is the value in col. (4). b A physically meaningful temperature cannot be derived when F 60 > (5=3) 4 F 100 ( ¼ 1). blue and red wings, respectively. Ho et al. (1982) analyzed the outflow gas with V < 16:9kms 1 and V > 5:1kms 1 for blue and red wings, respectively. The CO lines have prominent absorption at the line center. We have made channel maps to examine how the high-velocity gas is distributed from close to the line center to the line wing. The emission integrated within 6 to 3 kms 1 appears to show an additional component to the south of the red lobe. We checked the spectra and found that this velocity range coincides with the absorption. Thus, this component is probably not real but is due to absorption. We define the high-velocity range as V < 16 km s 1 for the blue wing and V > 3 kms 1 fortheredwingtoavoidtheselfabsorption. Our high-velocity ranges are closer to the line center than those of Rodríguez et al. (1980a). All the velocity ranges of bipolar outflows are indicated in the CO spectrum in panel (a) and the PV diagrams in panel (b) of the corresponding outflow figures. The outflow parameters were calculated following the procedure of Garden et al. (1991). Assuming LTE and optically thin emission in the CO gas, the column density of CO is given by N CO ¼ 1:1 ; 1013 (T ex þ 0:93) exp (5:5=T ex ) R T R (CO) dv cm 2 ; ½J (T ex ) J (T bg )Š½1 exp ( 11:1=T ex )Š where J (T ) is the Planck function and R T R (CO) dv is the integrated intensity of the high-velocity CO J ¼ 2 1 emission in units of K km s 1. A number of studies show that highvelocity gas is usually optically thick (Goldsmith et al. 1984; Snell et al. 1984; Shepherd & Churchwell 1996b). The effect of the optical depth on the mass estimation is complex. It is better to adopt a velocity-dependent (see Arce & Goodman 2001; Yu et al. 1999). So far the majority of authors have used constant values of for one or two velocity ranges, respectively. The average value for different sources or the entire line wing is usually around 4 (e.g., Snell et al. 1984; Goldsmith et al. 1984; Garden et al. 1991). Occasionally, ¼ 1 or 5 was used to see the effect of the optical depth on mass estimation (Shepherd & Churchwell 1996a; Chernin & Masson 1991). Thus, our optically thin assumption would produce a lower limit on the mass. The expression T ex is the excitation temperature in Kelvin. It changes with velocity and position, but using a proper constant T ex affects little the estimated total mass (Arce & Goodman 2002). Generally speaking, T ex in low-mass star-forming regions is about 10 K (Goldsmith et al. 1984), but about 30 K where high-mass stars form (Shepherd & Churchwell 1996a, 1996b). We adopted T d, which is calculated using IRAS flux density at 60 and 100 m and adopting a power law with index 2 for the dependence of dust opacity on the frequency. The dust temperature T d is usually larger than the gas temperature by as much as 10 K (Wu & Evans 1989). This difference would produce 30% uncertainty in the estimated column density. With the above procedure, the mass estimated is expected to be 3 times larger. For OMC 2, the value of T d is not physically meaningful, and the main-beam brightness temperature T mb ¼ T R = mb of thecolineisused.ifwetake½coš=½h 2 Š¼10 4 and the mean atomic weight of the gas g ¼ 1:36 to account for helium, the total mass M of the outflow gas can be estimated. The momentum P and energy E are proportional to R T(v)v dv and R T(v)v 2 dv, respectively, where v is the velocity of the gas relative to the cloud systemic velocity. The dynamical timescale t is estimated as R=V, wherer isassumedtobethemean distance from the peaks of the blueshifted and redshifted gas to the identified driving source. The expression V is the characteristic velocity derived from V ¼ P=M (Lada 1985). The mechanical luminosity is L ¼ E=t, and the driving force is F ¼ P=t. The mass-loss rate is Ṁ ¼ P=(tV w ), where the wind velocity V w is assumed to be 500 km s 1 (Marti et al. 1998). The outflow parameters we obtained for the six outflow sources are listed in Table 4. In the following, we discuss the sources individually. OMC 2. It is located in the north of the Trapezium OB star cluster. A cluster of low- to intermediate-mass pre mainsequence stars is seen in the near-infrared image. The compact near-infrared sources IRS 1, 2, 3, and 4N are the illuminating sources of the reflection nebulae (Rayner et al. 1989). IRS 4N (near the reference position, indicated as an open triangle in Fig. 1c) is associated with the H 2 S(1) line emission (Fischer et al. 1980). Wang (1997) also detected an H 2 jet originating

4 TABLE 3 Driving Source Identified Source 2MASS Source Other Near-Infrared Source IRAS Millimeter Continuum Centimeter Continuum References OMC a IRS 4N... 7, , 2, 3 GGD IRS , 3.6, 6 4, 5, 6 S ,8,9 IRAS a IRAS HW a , 10 IRAS b , 12 a There are two 2MASS sources within its error ellipse. b The deviation between the IRAS and the 2MASS source is References. (1) Rayner et al. 1989; (2) this paper; (3) Reipurth et al. 1999; (4) Yamashita et al. 1987; (5) Kurtz et al. 1994; (6) Rodríguez & Reipurth 1989; (7) Wang 1997; (8) Felli & Harten 1981; (9) Felli et al. 1978; (10) Hughes & Wouterloot 1984; (11) Molinari et al. 1998; (12) Gregory & Condon Fig. 1a Fig. 1b Fig. Fig. 1d 1. Figures for OMC 2. (a) Spectra. The left and right vertical lines in the top panel indicate the beginning of the blue and red wings, respectively. The bottom panel shows the line velocity shifts. (b) PV diagram. Contour levels are 0.5 to 2.0 K by 0.5 K, 4.0 to 12.0 K by 2.0 K, and 16.0 to 36.0 K by 4.0 K; the vertical lines show the same velocity values as those in (a). (c) Contour map of the wing emissions ( line-center velocity shifts corrected), integrated over about 7.5 to 7.5 km s 1 for the blue wing (solid line) and about km s 1 for the red wing (dashed line), respectively. The cross denotes the reference position for the relative coordinates (IRS 4). The filled triangles mark the locations of IRAS sources. Small crosses are the observed points. Contours begin from 17.0 K km s 1 with increasing steps of 4.2 K km s 1 for the blue wing and begin from 11.7 K km s 1 with increasing steps of 2.9 K km s 1 for the red wing. The H 2 jet detected by Wang (1997) was superposed on our outflow lobes. (d ) VLA map at 7 mm continuum emission. Contours are 3, 2, 1, 1, 2, 3, 4, 5, 6, 7, 8, and 10 ; 0.6 mjy beam 1. The circle at the bottom left corner marks the synthesized beam.

5 334 WU ET AL. Vol. 129 Fig. 1c Fig. 1d from IRAS 4N. IRAS , labeled as IRAS 1 in the map (Fig. 1c), is located near IRS 3. Another IRAS source, IRAS (labeled as IRAS 2 in the map), is 3A5north and located at the edge of the map. Two H 2 O masers were detected in this region (Tofani et al. 1995). Fischer et al. (1985) detected an outflow in the CO J ¼ 1 0 line, and the blue and the red lobes overlap considerably. We present in the top panel of Figure 1a acoj ¼ 2 1 spectrum taken from near the position of IRS 4N. The line exhibits a large velocity extent ( 7.5 to 26.5 km s 1, measured at the 0.2 K level). The spectra in the bottom panels are taken from different positions and show a shift in line-center velocities. A PV diagram is shown in Figure 1b. The cut is along the outflow axis and through the position of IRS 4N. The boundary between the high-velocity gas and the ambient gas and the different spatial distribution of the blue wing and the red wing can be clearly seen. Figure 1c shows contours of the blue- and the redshifted gas emission. The H 2 jet detected by Wang (1997) was superposed on our outflow lobes. From the near-infrared image, one can see clearly a bow shock pointing to the northeast direction (Wang 1997). It is consistent with the morphology of the molecular outflow. The near-infrared source IRS 4N is identified to be the driving source for the OMC 2 outflow by Rayner et al. (1989). It is located nearly on the axis of the CO outflow and the H 2 jet (Wang 1997). To further investigate the driving source, we observed 7 mm continuum emission with VLA. We found that a compact structure almost coincides with IRS 4N. The 7 mm source corresponds to VLA 11 (Reipurth et al. 1999). It shows the existence of thermal emission there and confirms IRAS 4N as the driving source. We use the flux density of the IRAS source to estimate the bolometric luminosity of the driving source. GGD 27. This is an infrared reflection nebulosity (Hartigan & Lada 1985). There are at least three stellar sources in this region, of which IRS 2 is the illuminating source (Yamashita et al. 1987). There is only one IRAS source, IRAS , within the map, which corresponds to IRS 2. Awater maser was found near GGD 27 by Rodríguez et al. (1980b). A bipolar molecular outflow was detected in CO J ¼ 1 0 by Yamashita et al. (1989). Our CO J ¼ 2 1 results are shown in Figure 2. The total velocity width at zero intensity for CO J ¼ 2 1at( 0A24, 0A24) is 46 (from 6 to 40) km s 1, which is larger than that of CO J ¼ 1 0 at the same rms level. In the bottom panel of Figure 2a, we present two spectra showing shifts in line velocity. Figure 2b presents the PV diagram. The boundary between the outflow and the core material is clear. Comparing with the diagram in Yamashita et al. (1989), one can see that the CO J ¼ 2 1 line has wings that extend further in velocity than those of the CO J ¼ 1 0 line. The spatial extent of the wing emission is smaller. The distribution of the blue- and redshifted gas is presented in Figure 2c. The total mass in the flow obtained in this paper is 27 M, much less than the 460 M obtained by Yamashita et al. (1989) from CO J ¼ 1 0. In estimating outflow mass from CO J ¼ 1 0, Yamashita et al. used optical depths of for each lobe. IRS 2 and IRAS are located near the flow axis and are shown as open and filled triangles, respectively. The IRAS source has colors satisfying the criteria of the ultracompact (UC) H ii regions (Wood & Churchwell 1989) and a large FIR luminosity. At its position, continuum emission at 6 cm was detected by Rodríguez & Reipurth (1989). The image of the 6 cm continuum was superposed on the flow. Kurtz et al. (1994) imaged the UC H ii region at 2.0 and 3.6 cm and reported a radio source G at (B1950) ¼ 18 h 16 m 13 s, (B1950) ¼ B7. These observations confirm that IRS 2 is the driving source of the outflow. The corresponding 2MASS source is at (B1950) ¼ 18 h 16 m 13: s 17, (B1950) ¼ B08 and is , , and mag at the J, H, andk s bands, respectively. S146. This is an H ii region with weak H emission (Blair et al. 1975). H 2 O maser emission with multiple components has been detected (Blair et al. 1978; Henkel et al. 1986). Both the optical and the 6 cm images show a bipolar structure (Felli & Harten 1981). Mapping in the CO J ¼ 1 0 line shows a bipolar outflow (Yang & Wu 1998). The CO J ¼ 2 1spectrain Figure 3 show asymmetric line profiles and broad wings (see the top panels of Fig. 3a). The PV diagram with a cut along the outflow axis in the north-south direction (Fig. 3b) shows the high-velocity outflow. The contours of the wing emission in Figure 3c show that the blue lobe and the red lobe overlap. The peaks of the two lobes are about apart. The total line width in CO J ¼ 2 1( 60 to 35 km s 1 )islargerthanthatof CO J ¼ 1 0( 57.5 to 41.5 km s 1 ; Yang & Wu 1998). Wang (1997) imaged this region in the J, H, andk bands.

6 No. 1, 2005 BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS 335 Fig. 2. Figures for GGD 27. (a) Spectra. The left and right vertical lines in the top panel indicate the beginning of the blue and red wings, respectively. The bottom panel shows the line velocity shifts. (b) PV diagram. Contour levels are from 0.5 to 1.5 K by 0.25 K, 2.1 to 4.8 K by 0.5 K, and 6.2 to 14.4 K by 1.3 K; the vertical lines show the same velocity values as those in (a). (c) Contour map of the wing emissions, integrated over 7.4 to 8.5 km s 1 for the blue wing (solid line) and km s 1 for the red wing (dashed line), respectively. The filled triangle marks the IRAS source position (the reference position). Contours begin from 16.1 K km s 1 with increasing steps of 4.0 K km s 1 for the blue wing and begin from 18.5 K km s 1 with increasing steps of 4.5 K km s 1 for the red wing. The thick lines present the 6 cm image of the excitation source (Rodriguez & Reipurth 1989). Number 132 in Wang s near-infrared source list [at ( B1950) ¼ 22 h 47 m 30: s 4, (B1950) ¼ B5] is close to the flow axis and the red peak and is indicated as an open triangle in Figure 3c. It is also the brightest source (12.4, 11.2, and 10.1 mag at the J, H, and K bands, respectively) near the flow peak. The corresponding 2MASS source is at (B1950) ¼ 22 h 47 m 30: s 49, (B1950) ¼ B13 and with , , and mag at the J, H, andk s bands, respectively. The positions and near-infrared magnitudes detected by Wang (1997) and 2MASS are consistent. The IR source coincides with IRAS and the 6 cm emission peak (Felli et al. 1978). The L bol of the IRAS source is 6:2 ; 10 4 L at a

7 336 WU ET AL. Vol. 129 Fig. 3. Figures for S146. (a) Spectra. The left and right vertical lines in the top panel indicate the beginning of the blue and red wings, respectively. The bottom panel shows the line velocity shifts. (b) PV diagram. Contour levels are from 0.5 to 1.5 K by 0.5 K, 2.0 to 5.0 K by 1.5 K, and 7.5 to 21 K by 2 K; the vertical lines show the same velocity values as those in (a). (c) Contour map of the wing emissions, integrated over 60.0 to 52.0 km s 1 for the blue wing (solid line) and 45.0 to 35.0 km s 1 for the red wing (dashed line), respectively. Contours begin at 6.5 K km s 1 with increasing steps of 2.0 K km s 1 for the blue wing and begin at 7.5 K km s 1 with increasing steps of 2.4 K km s 1 for the red wing. The open triangle denotes near-infrared source number 132 in the list of Wang (1997). distance of 5.2 kpc. A star with spectral type O6.5 or earlier is needed to account for the ionization of the H ii region and the infrared radiation ( Felli & Harten 1981). This source is likely to be responsible for the driving of the outflow. IRAS This source was named (Comoretto et al. 1990). It is located beyond the solar circle. Strong water maser emission has been detected with multiple velocity components (Wouterloot & Walmsley 1986; Comoretto et al. 1990). VLA observations show that there are three masers in this region (Jenness et al. 1995). The CO J ¼ 1 0 line shows two components at 51.0 and 9.29 km s 1 (Wouterloot & Brand 1989). Wu et al. (1998a) measured the CO J ¼ 1 0linewidthat

8 No. 1, 2005 BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS 337 Fig. 4. Figures for IRAS (a) Spectrum. The left and right vertical lines indicate the beginning of the blue and red wings, respectively. (b) PV diagram. Contour levels are 0.5 to 1.5 K by 0.5 K, 2.0 to 5.0 K by 1.5 K, and 7.5 to 21 K by 2 K; the vertical lines show the same velocity values as in (a). (c) Contour map of the wing emissions, integrated over 69.3 to 53.7 km s 1 for the blue wing (solid line)and 47.0 to 31.7 km s 1 for the red wing (dashed line), respectively. Contours begin at 12.7 K km s 1 with increasing steps of 3.1 K km s 1 for the blue wing and begin at 11.5 K km s 1 with increasing steps of 2.9 K km s 1 for the red wing. The filled triangle marks the locations of the IRAS source. Small crosses are the observed points. The open triangle denotes the near-infrared source S4 (number 83) detected by Wang (1997). zero intensity to be 15.1 km s 1. The CO J ¼ 2 1 spectrum shown in Figure 4a is taken at (;) ¼ (0A20; 0A24) from the reference position. The total line width at zero intensity is about 35 km s 1. The PV diagram (Fig. 4b) with a cut along the north-south direction shows high-velocity wings. The contour map of the blue- and redshifted gas in Figure 4c shows that the two lobes largely overlap. Two of the three masers are also located close to the flow peak position (Jenness et al. 1995). Wang (1997) imaged the source in the near-infrared J, H, and K bands and found a cluster within the cloud core. The one labeled S4 (number 83 in Wang s near-infrared photometry list) resides at (B1950) ¼ 22 h 50 m 38: s 85, (B1950) ¼ B6, close to the peak of the outflow, which is presented as an open triangle in Figure 4c. It has magnitudes of 17.7, 14.5, and 11.7 in the J, H, and K bands, respectively (Wang 1997). It corresponds to a 2MASS source, , at (B1950) ¼ 22 h 50 m 38: s 83, (B1950) ¼ B49 with , , and mag in the J, H, and K s bands, respectively. S4 is closer to the flow center and is likely to be the driving source. It also coincides with IRAS The IRAS source has a luminosity of 1:5 ; 10 4 L at a distance of 5.1 kpc. Its colors satisfy the UC H ii region criteria (Wood & Churchwell 1989). The bolometric luminosity of IRAS is taken as the luminosity of the driving source in the following analysis. IRAS The outflow in Cep A is one of the first outflows discovered (Rodríguez et al. 1980a; Ho et al. 1982). IthasbeenmappedinboththeCOJ ¼ 1 0 line and later in the CO J ¼ 3 2 line (Narayanan & Walker 1996; Choi et al. 1993). A strong water maser exists in this source (Wu et al. 1999; Comoretto et al and references therein). The results in the CO J ¼ 2 1 line are shown in Figure 5. The spectra in Figure 5a are taken from three positions. The blue wing extends to 40 km s 1 and the red one to 21.6 km s 1. It appears that the line profiles are relatively complex and there are dips in the line center. Based on the variation of the line profiles from the edge to the center of the map, we think that the peak at about 11.5 km s 1 is the counterpart of the component in the edge region. The peak agrees well with the CS J ¼ 7 6observation ( 12 km s 1 ; Narayanan & Walker 1996) and represents the ambient gas. The velocity variations of this peak contain an east-west component and a north-south component. The velocity increases about 2 km s 1 in a region of 2A5 along the northeast to southwest direction (shown in Fig. 5a). The velocity shifts agree in general with the CS line measurements of Narayanan & Walker (1996). Figure 5b shows the PV diagram along the outflow axis (east-west direction). The spatial separation of the blue and the red wing is apparent. Figure 5c presents the contours of the high-velocity gas emissions. The

9 Fig. 5b Fig. 5c Fig. 5a Fig. 5d Fig. Fig. 5d 5. Figures for IRAS (a) Spectra. The left and right vertical lines in the top panel indicate the beginning of the blue and red wings, respectively; the bottom panel shows the line velocity shifts. (b) PV diagram. Contour levels are 0.7, 1.0, 1.5, 2.0, 4.0, 6.0, 8.0, 11.0, 13.0, and 14 K; the vertical lines show the same velocity values as those in (a). (c; left) Contour map of the wing emissions, integrated over 40.0 to 16.0 km s 1 for the blue wing (solid line) and 3.0 to 21.6 km s 1 for the red wing (dashed line), respectively. Contours begin at 27.2 K km s 1 with increasing steps of 6.8 K km s 1 for the blue wing and begin at 20.3 K km s 1 with increasing steps of 5.1 K km s 1 for the red wing. (Right) The 6 cm continuum image of HW 2 ( Hughes & Wouterloot 1984). (d ) Channel maps of velocity ranges ( 6, 3) km s 1 (left) and( 3, 0) km s 1 (right). Contour levels are from 10 to 26 K by 4 K.

10 BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS 339 Fig. 6. Figures for IRAS (a) Spectra. The left and right vertical lines indicate the beginning of the blue and red wings, respectively. (b) PV diagram. Contour levels are 0.25 K, 0.5 to 1.5 K by 0.5 K, 2.0 to 5.0 K by 1.5 K, and 7.5 to 21 K by 2 K; the vertical lines show the same velocity values as those in (a). (c) Contour map of the wing emissions, integrated over 61.7 and 53.4 km s 1 for the blue wing (solid line) and 46.1 to 41.8 km s 1 for the red wing (dashed line), respectively. Contours begin at 4.6 K km s 1 with increasing steps of 1.0 K km s 1 for the blue wing and begin at 7.5 K km s 1 with increasing steps of 1.2 K km s 1 for the red wing. The 3.4 mm emission image is superposed on the contours as a shaded region (Molinari et al. 1998). outflow mass we computed is similar to that of CO J ¼ 1 0, considering the different ratio of 12 CO=H 2 (Rodríguez et al. 1980a). HW 2 was identified as the driving source of the outflow (Hughes & Wouterloot 1984; Narayanan & Walker 1996; Torrelles et al. 1998). Its 6 cm continuum image is indicatedintheoutflowmapinfigure5c (Hughes & Wouterloot 1984). The 2MASS source [(B1950) ¼ 22 h 54 m 20: s 873, (B1950) ¼ B23] coincides with HW 2. Its magnitudes in the J, H, andk s bands are , , and 8.640, respectively. HW 2 is located between the peaks of the blue and the red lobes and on the outflow axis, suggesting that it drives the outflow. IRAS This H 2 O maser source (Comoretto et al. 1990) was found to have an asymmetric line profile and

11 340 WU ET AL. Vol. 129 Fig. 7. Figures for IRAS (a) Spectra. (b) PV diagram. Levels are at 0.5, 1.0, 1.5, 2.0, 3.5, 5.0, 7.5, 10.5, 13.5,...,37.5Kkms 1.(c) Contour map of the blue lobe, integrated over 90.8 to 98.0 km s 1. Contour levels are from 12.5 to 19.3 K km s 1 by 1.7 K km s 1. red wings in the CO J ¼ 1 0 line (Wouterloot & Brand 1989). Yang & Wu (2000) mapped the CO J ¼ 1 0 line and identified an outflow associated with a high-mass YSO. Molinari et al. (1998) observed HCO + and SiO, and an outflow was detected. They also measured emission at 15 m and 3.4 mm. The spectra of CO J ¼ 2 1inFigure6a show that the CO J ¼ 2 1 lines also have asymmetric profiles and wings. The total width ( 61.7 to 41.8 km s 1 )islargerthanthatofcoj ¼ 1 0 ( 57.9 to 41.3 km s 1 ). Molinari et al. (1998) set V < 53 km s 1 asthebluewingandv > 47 km s 1 as the red wing for the SiO spectra. Our high-velocity ranges are defined by the PV diagram along the outflow axis (north-south direction), which is presented in Figure 6b. The PV diagram clearly shows no high-velocity gas less than 46 km s 1 at the redshifted side. We take the high-velocity ranges as V < 53 km s 1 and V > 46 km s 1 for the blue and red wings, respectively. In both the outflow contours of SiO (Molinari et al. 1998) and CO J ¼ 2 1 of this work, the red lobe is rather diffuse, which is due to the rather small velocity gradient at 47 km s 1 and also at 46 km s 1. The map of the integrated emission of the blue and red wings (Fig. 6c) shows that the two lobes mostly overlap, which may indicate a nearly pole-on outflow. The spatial extent of the CO J ¼ 2 1 outflow is smaller than that of CO J ¼ 1 0 (Yang & Wu 2000). The center of the outflow lobes is associated with the peak of the 3.4 mm continuum contours, of which the image is shown in Figure 6c. This image shows that the 3.4 mm compact source, which corresponds to a very young stellar object, probably drives the outflow. The location of IRAS (see Table 2) is offset about from the millimeter emission peak [(B1950) ¼ 23 h 38 m 31: s 4, (B1950) ¼ (Molinari et al. 1998)]. At the position of (B1950) ¼ 23 h 38 m 28: s 5 1: s 2, (B1950) ¼ , a continuum source was detected at 4.85 GHz (Gregory & Condon 1991), which is within the IRAS error ellipse (13 00 ; 3 00 ). Therefore, the

12 No. 1, 2005 BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS 341 Fig. 8. Figures for G (a) Spectra. (b) PV diagram. Levels are at 0.5, 1.0, 1.5, 2.0, 3.5, 5.0, 7.5, 11.5, 15.5,...,39.5Kkms 1.(c) Contour map of the blue lobe, integrated over 17.7 to 25.0 km s 1. Contours begin at 6.8 K km s 1 with increasing steps of 3.4 K km s 1. Class 0 object mentioned by Molinari et al. (1998) may not be the dominant IRAS source but a second, weaker object within the IRAS beam. We take the bolometric luminosity of the IRAS source as the luminosity of the driving source Sources in which Bipolar Outflows Cannot Be Identified IRAS This is an H ii region corresponding to the radio source G H 2 O maser emission was detected at 102 km s 1 by Churchwell et al. (1990). Observations in CO J ¼ 1 0 showed a total line width at zero intensity of 34.5 km s 1 (Wu et al. 1998a). The CO J ¼ 2 1 map in this paper covers a region of 5A6 ; 2A8. The line profiles are asymmetric and broad. Two spectra taken from (;) ¼ (0A24; 0A24) and ( 0A73, 2A66) are shown in Figure 7a. Thespectrum in the top panel seems to show both blue and red wings. However, in the spectrum in the bottom panel, we can see a minor peak at 114 km s 1, about 10 km s 1 from the main component. The emission of this red peak remains constant throughout the map and represents a component of extended emission. Meanwhile, the blue wing extends more than 12 km s 1 from the line center, indicative of high-velocity wings. Figure 7b shows the PV diagram. We can see again the extended cloud components at the redshifted side. The contours of the blueshifted emission is mainly distributed in the west and the north of the central source ( Fig. 7c). The redshifted emission cannot be identified because of the confusion by the other component. G It is a strong H 2 O maser source (Comoretto et al and references therein). Dent et al. (1984) mapped the H ii region at 6 cm. The dominant source in the region, IRAS , has color indexes higher than the criteria for H ii regions and a luminosity of 1:6 ; 10 4 L at a distance of 2 kpc. AweakerIRAS source, IRAS , is located 3 0 north of the reference position. Matthews et al. (1984) observed the region in the HCO + J ¼ 1 0andJ ¼ 3 2 lines. They reported a bipolar outflow in the northern part of the source around (B1950) ¼ 18 h 55 m 41: s 0, (B1950) ¼ Dentetal. (1985) mapped the source in CO J ¼ 1 0andCSJ ¼ 2 1. They obtained a CO bipolar outflow and a CS core. Our CO J ¼ 2 1 map was made with an offset position (B1950) ¼ 19 h 07 m 14: s 7, (B1950) ¼ , which is cleaner than the one used by Dent et al. (1985). The size of the map is larger than the HCO + and the CO J ¼ 1 0maps.TheCOJ ¼ 2 1 results show that there are more than two cloud components. Four CO J ¼ 2 1 spectra taken from different positions are shown in Figure 8a, of which one is from the central region near IRAS and the others are from the edge region. Lines are very broad in the central region around IRAS There is a strong self-absorption at the line center. This agrees

13 342 WU ET AL. Vol. 129 Fig. 9. Figures for IRAS (a) Spectra. (b) PV diagram. Contour levels are at 0.5, 1.0, 1.5, 2.0, 3.5, 5.0, 7.5, 9.5, 11.5,...,19.5Kkms 1. well with previous observations. In the entire region, there are perhaps five components at 14.2, 12.8, 32.7, 44.6, and 53.2 km s 1, respectively. The PV diagram (Fig. 8b) shows weak velocity components at about 10, 40, and 50 km s 1. Excluding the component at 12.8 km s 1, we made contours for the blueshifted wing of the main component. The range of integration is from 17.7 to 25 km s 1. The map is shown in Figure 8c. The blue lobe is consistent with that of Dent et al. (1985). The red lobe cannot be determined because of the additional component at 44.5 km s 1. In the CO J ¼ 1 0 results of Dent et al., the red wing is adopted as km s 1, which includes the 44.5 km s 1 component. IRAS This source is located in the west of W51 at a distance of 6.8 kpc (Genzel & Downes 1977). It is a strong IRAS source with a luminosity of 7:7 ; 10 5 L.ItsIRAS color indexes satisfy the criteria of UC H ii regions. H 2 Omaser Fig. 10. Figures for IRAS (a) Spectra.(b) PV diagram. Contour levels are at 0.5, 1.0, 1.5, 2.0, 3.5, 5.0, 7.5, 9.5, 11.5,...,19.5Kkms 1.

14 No. 1, 2005 BIPOLAR OUTFLOWS IN MASSIVE STAR-FORMING REGIONS 343 Fig. 11. Figures for IRAS (a) Spectrum. (b) PV diagram. Levels are at 0.5, 1.0, 1.5, 2.0, 3.5, 5.0, 7.5, 9.5, 11.5,..., 39.5 K km s 1. emission at about 52.0 km s 1 with multiple components has been detected at (;) ¼ (17 00 ; ) southwest of IRAS (Genzel & Downes 1977; Comoretto et al. 1990). Two CO J ¼ 2 1 spectra are shown in Figure 9a. Thepeak emission is at 53 km s 1. The spectrum in the top panel has blue and broad red wings. The one in the bottom panel presents two additional components at about 58 and 66 km s 1.Thereis likely the high-velocity blue wing, but the red wing is confused by these two velocity components. IRAS This is the only IRAS source within the mapped region. The IRAS color indexes satisfy the criteria of UC H ii regions. Wouterloot & Brand (1989) observed it in the CO J ¼ 1 0 line and found two components at 53.3 and 47.9 km s 1.TheCOJ¼1 0 lines observed by Wu et al. (1998b) have a width at zero intensity of about 26 km s 1.Two CO J ¼ 2 1 spectra are shown in Figure 10a. The one taken from the central region at (;) ¼ ( 0A24; 0A24) shows a velocity range from 58.5 to 43.0 km s 1 at zero intensity. The spectrum taken from the northern edge at (;) ¼ (0A73; 2A66) shows two components at 52.7 and 47.5 km s 1,respectively. The velocities of the two components agree with those of the CO J ¼ 1 0( 53.3 and km s 1 ). However, the relative intensities are different. (The beam sizes of the telescopes used in the two observations are about the same.) In the central region, the two CO J ¼ 1 0 components are 15.6 and 13.9 K, respectively (Wouterloot & Brand 1989), whereas for the CO J ¼ 2 1, the TA of the two components are about 23.5 and 2.5 K, respectively. In addition, in the J ¼ 2 1 line, the component at 52.7 km s 1 decreases rapidly toward the edge, but the component at 47.5 km s 1 remains constant. This may suggest that the excitation conditions are different for these two components. These two components can also be seen in the PV diagram ( Fig. 10b). IRAS : This is the only IRAS source within the mapped region. The IRAS color indexes satisfy the criteria of H ii regions. VLA observations show strong emission at 3.6 and 2.0 cm wavelengths (Kurtz et al. 1994). The center of the radio source G is at (B1950) ¼ 23 h 13 m 21: s 25, (B1950) ¼ B5, which coincides with the IRAS source. The CO J ¼ 1 0 line shows a large line width (24.3 km s 1 at zero intensity) and broad wings (Wu et al. 1998b), similar to the CO J ¼ 2 1lineinFigure11a. Figure11b shows its PV diagram. 4. DISCUSSION 4.1. Comparison between Results of the CO J = 2 1 and the CO J = 1 0 Lines The velocity extents of line wings differ between the CO J ¼ 1 0andJ ¼ 2 1lines.ForOMC2,theCOJ ¼ 2 1velocity ranges from 7.5 km s 1 to around 25 km s 1, whereas it is from 0 to 20 km s 1 for CO J ¼ 1 0 (Fischer et al. 1985). In GGD 27 and IRAS , the total widths in the CO 2 1 line are 48 (from 8 to 40) km s 1 and61.6(from 40 to 21.6) km s 1, respectively. These are much larger than the widths in the CO J ¼ 1 0 line (6 19 km s 1 and 36 to 14 km s 1, respectively; Yamashita et al. 1989; Rodríguez et al. 1980a). The total CO J ¼ 2 1 widths for other bipolar outflows (S146 and IRAS ) are also larger than those of CO J ¼ 1 0by5 10kms 1 (see Yang & Wu 1998, 2000). The TABLE 4 Parameters of Outflows Source Size ( pc) M (M ) P (M km s 1 ) E (10 45 ergs) t (10 4 yr) L (L ) F (10 3 M km s 1 yr 1 ) Ṁ (10 5 M yr 1 ) OMC GGD S IRAS IRAS IRAS

15 344 WU ET AL. Vol. 129 Fig. 12. Correlations of outflow parameters: (a) mechanical luminosity vs. bolometric luminosity; (b) driving force vs. bolometric luminosity. Sources from Cabrit & Bertout (1992) are plotted as stars, of which those for massive sources (with L bol > 1000 L ) are put in circles and those for intermediate-mass ones (with L bol between 100 and 1000 L ) in squares. Sources from Shepherd & Churchwell (1996b) are plotted as open circles. Sources from this work are plotted as triangles, of which the one for OMC 2 is plotted in a square. The uncertainties in the parameters are shown in the bottom right corner. spatial extent of the CO 2 1 outflow for OMC 2 agrees in general with the result of the CO J ¼ 1 0 outflow (Fischer et al. 1985) but is narrower in the northwest-southeast direction. For GGD 27, the outflow regions measured with the CO J ¼ 2 1 line are smaller than those of the CO J ¼ 1 0 line with the same integration range. For IRAS , the halfpower contour of the CO J ¼ 2 1 outflow is smaller than that of CO J ¼ 1 0, even considering the different spatial resolutions and the different velocity intervals of the cores (Rodríguez et al. 1980a). In S146 and IRAS , the outflow regions are smaller than those measured in the CO J ¼ 1 0 line. The mass of the CO J ¼ 2 1 outflows may not be compared directly with those of the CO J ¼ 1 0 line for several reasons: the ratio of ½H 2 Š=½ 12 COŠ used, the excitation temperature adopted, the line intensity measured, the area of the flow region determined, and the velocity range of the wings chosen. For those in which masses can be compared, the masses of the CO J ¼ 2 1 outflows are usually smaller than those of the CO J ¼ 1 0line. The broader line wings seen in the CO J ¼ 2 1 line suggest that outflows in massive star-forming regions arise from relatively warm gas. Calculations show that under LTE conditions, the CO J ¼ 2 1 line is expected to be stronger than the CO J ¼ 1 0 line for gas temperatures of more than 20 K. We found that the peak brightness temperature exceeds 20 K in regions observed here. In addition to the benefit of achieving higher angular resolution compared to the CO J ¼ 1 0 line, the CO J ¼ 2 1 line is more sensitive to warm outflows. These warm gases are localized to the massive YSOs in massive star-forming regions Outflow Parameters We analyzed the relationship between the outflow parameters and the properties of the driving source identified as above. The outflow parameters are listed in Table 4. It appears that the outflows in high-mass star-forming regions carry significantly larger amount of masses and momenta than those of low-mass outflows. Since high-mass stars typically form in a cluster with members of lower masses, there have been suggestions that high-mass outflows are driven by low-mass YSOs in the cluster. However, the masses and momenta in high-mass outflows are at least an order of magnitude larger than in typical low-mass outflows (Cabrit & Bertout 1992; Myers et al. 1988). The sensitivity and spatial resolution of our observations are not high enough to detect single molecular outflows driven by individual low-mass stars. Assuming a Salpeter initial mass function and a scaling law of outflow mass-loss rate and the luminosity of the star (see Tan & McKee 2002), the molecular outflow from low-mass stars can exceed the mass-loss rate from the massive star alone and dominate the flow in the cluster. Observationally, this has not been confirmed. Since the axes of low-mass outflows are not necessarily aligned in the cluster, they contribute to the blue- and red-shifted gas randomly. Thus, one should expect to observe in many outflows in massive star-forming regions that the blue- and redshifted lobes overlap spatially, if low-mass outflows follow the scaling law. This is not what has been observed: at an spatial resolution of 30 00, we resolve the outflows toward S146 and IRAS At resolution, Beuther et al. (2002a) shows toward a larger sample that outflow lobes from massive star-forming regions are spatially well separated. In addition, higher angular resolution data from radio interferometers (e.g., Shepherd et al. 1998, 2000; Beuther et al. 2002b, 2003; Su et al. 2004) also show that these outflows are associated with the more massive members in the cluster. Thus, it is most likely that the high-mass YSOs drive the high-mass outflows. To further analyze the relationship between the outflow parameters and the driving sources, we plotted the correlation in Figure 12. For comparison, we also included sources from Cabrit & Bertout (1992) and Shepherd & Churchwell (1996b). Figures 12a and 12b show the correlation between the CO outflow luminosity L CO and L bol, and the mechanical force and L bol, respectively. The bolometric luminosity of the central source is calculated with the IRAS fluxes. They are close to the results derived from other infrared observations (see Wu et al. 2004). Mechanical luminosity and the force required by the outflow have most of their uncertainty from the mass determination.